Collect end-to-end analytics data in a single database

[Reddi2]

The next step in building end-to-end analytics will be the creation of a single database that will store and process information from various sources (advertising sources, sources transmitting statistics). This database can be Google BigQuery, ClickHouse, various DBMS options (PostgreSQL, MySQL, and others), or S3. It will not be possible to fully perform the task of preliminary data processing in Google Analytics, since the service does not have such functionality.

Combine customer data from different sources
At this stage, using analytical tools (for example, those included in the Google BigQuery functionality), we mutually complement data about users and their interactions with our marketing channels and sales department. As a result, we can understand who and how interacted with us, segment potential and current clients, determine the ideal client profile and see effective and ineffective channels of interaction with them.

Decide on an attribution model
A few words about efficiency. Ultimately, the efficiency of a the benefit of using our database channel is not an absolute truth, but is largely determined by the attribution model you choose — a rule or set of rules that determine the contribution of the interaction channel to the conversion. The choice of attribution model should be given special attention, since it determines which channels you decide to continue working with and which ones you don’t. The most well-known models are first click (only the first interaction is taken into account), last click (only the last interaction is taken into account), post-click conversion (attribution window after interaction) and the linear model (the weights of all interaction channels are distributed evenly).